[
https://issues.apache.org/jira/browse/ZOOKEEPER-1777?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Germán Blanco updated ZOOKEEPER-1777:
-------------------------------------
Attachment: logs_trunk.tar.gz
It does occur in trunk also. Logs attached in file logs_trunk.tar.gz.
However I did see the TRUNC used for synchronization a couple of times, and
also a message of being unable to send TRUNC because of different epochs and
sending snapshot instead. So it was a bit harder to reproduced.
This is the data in server A:
[Fbc, Cbc, 6a, 4a, 7bc, 5a, 8bc, 3, 2, 1, 9a, 9bc, 7a, 8a, zookeeper, Abc, 6bc,
Bbc, Dbc, Ebc]
This is the data in server B:
[Fbc, Cbc, 7bc, 5bc, 8bc, 4bc, 3, 2, 1, 9bc, zookeeper, Abc, 6bc, Bbc, Dbc, Ebc]
I am working in the patch that sends information about the last transaction
from the learner to the leader. That means that synchronization via snapshot
will only happen when this problem occurs. Personally I don't see any other way
to solve this, please tell me if you do.
> Missing ephemeral nodes in one of the members of the ensemble
> -------------------------------------------------------------
>
> Key: ZOOKEEPER-1777
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1777
> Project: ZooKeeper
> Issue Type: Bug
> Components: quorum
> Affects Versions: 3.4.5
> Environment: Linux, Java 1.7
> Reporter: Germán Blanco
> Assignee: Germán Blanco
> Priority: Blocker
> Fix For: 3.4.6, 3.5.0
>
> Attachments: logs_trunk.tar.gz, snaps.tar, ZOOKEEPER-1777.tar.gz
>
>
> In a 3-servers ensemble, one of the followers doesn't see part of the
> ephemeral nodes that are present in the leader and the other follower.
> The 8 missing nodes in "the follower that is not ok" were created in the end
> of epoch 1, the ensemble is running in epoch 2.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
